Introduction: the challenges of identifying root causes in construction and engineering disputes
How long does it take to count to 2 million? If one were to count for eight hours a day, it would take 70 days to reach 2 million. Now, imagine reviewing 2 million documents manually, logging relevant information into Excel. That would take over 20 years! Why is this relevant? Because in construction and engineering disputes, millions of records hold the answers, yet traditional approaches make extracting them a near-impossible feat. Construction and engineering disputes often hinge on establishing a clear cause and effect relationship between delays, cost overruns, and technical issues. However, the sheer volume of project records: emails, reports, drawings, schedules, and quality assurance documentation, poses a significant challenge.
Author: Vladimir Milovanovic, Chief Executive Officer, LUPA Technology
The critical evidence is often buried in vast amounts of unstructured data making traditional methods of dispute resolution slow, costly, and prone to subjective interpretations. Historically, experts have relied on high-level documents such as formal letters and nonconformance reports (NCRs). However, these documents are often shaped by contractual positions rather than an objective representation of events.
Meanwhile, raw site communications where the real project story unfolds, remain largely untapped due to the overwhelming effort required to process them manually. How can dispute resolution be improved to establish a precise and verifiable cause-effect relationship? The answer lies in combining structured data management, advanced analytics, and AI-driven insights to extract and validate key facts efficiently.
The traditional approach: limitations and challenges
Proving cause and effect in construction and engineering disputes has always been an uphill battle.
Traditional dispute resolution methods rely on:
- Manual document review, requiring months or even years to process large data sets.
- Selective and incomplete evidence, often shaped by legal and commercial interests rather than factual accuracy.
- Expert analysis based on limited records, increasing the risk of biased conclusions and weak argumentation, built on selective records rather than a complete dataset. This is particularly relevant in dispute resolution proceedings where actual project events matter, and where an accurate storyline with substantiation is key to success.
Case study: the pitfalls of a traditional approach
In a project delay claim worth $100 million, a delay expert was tasked with reviewing over 2 million emails, written in both English and Dutch. Without automation, language barriers and data volume made a comprehensive review impractical. The conventional approach: translated summaries, keyword searches, and selective document sampling would have taken years and left critical gaps in the evidence. The inefficiencies of this approach result in escalated costs, drawn-out legal battles, and increased uncertainty in dispute outcomes.
A new approach: ai and data-driven cause-effect mapping
To form an opinion, two fundamental elements are required:
- A clear narrative that accurately maps cause to effect.
- Substantiated claims backed by verifiable records. Advanced data-driven tools transformed this process. Using semantic search, automation, tagging, and summarisation, the expert identified 200 relevant emails per head of claim within just three hours - an impossible feat with traditional methods.
A powerful data management system, coupled with advanced construction analytics and AI insights, allowed the expert to:
- Label all documents and emails with specific predefined contexts.
- Generate real-time situational reports of project activities going back four years, analysing patterns and trends leading to the dispute.
- AI-driven summarisation of responsive documents to speed up review.
- Identify and validate responsive records, ensuring AI accuracy and reducing hallucinations.
- Exporting 200 key emails/documents for each head of claim, forming a clear and substantiated narrative.
Case study: enhancing dispute resolution with AI
In the previously mentioned case, semantic search and AI-assisted classification allowed the expert to pinpoint 200 highly relevant emails per claim within just three hours, a task that would have been impossible through manual review. This data-driven methodology not only reduced review time but also uncovered additional storylines that further substantiated the expert’s opinion.
Traditional vs. New Approach:
- Traditional: Experts sift through scattered records for months, to form their independent opinion which assists the court/tribunal.
- New Data-driven: AI pinpoints precise, relevant information in hours, forming a robust opinion.
From fragmented data to a coherent argument
Beyond efficiency, automation provided deeper insights, revealing five additional narratives that further substantiated the expert’s opinion.
Traditionally, construction disputes are hindered by:
- Selective or biased interpretation of evidence – Stakeholders often present documents that align with their position while overlooking contradictory records.
- Incomplete/missing documentation – Critical emails, site logs, and reports may remain undiscovered due to volume and complexity.
By utilising concept-based classification and AI-driven summarisation, the expert developed a multi-faceted opinion, demonstrating that the delay resulted from multiple interconnected factors.
Meanwhile, the opposing party was faced with increasing challenges.
Indicators of a Weakening Position:
- Key personnel were dismissed mid-process, disrupting the continuity of their argument.
- New delay experts were introduced late in the proceedings, distancing themselves from prior assessments.
- Supporting documentation became increasingly inconsistent, exposing critical gaps in their case.
Leveraging data visualisation for dispute resolution
Clear documentation and visual representation of evidence are essential in complex disputes, even more so considering that 30% of design changes are not clouded on drawings. With AI-powered document management, thousands of drawings can be analysed, summarising all design changes automatically.
AI-powered tools enable this by:
- Identifying undocumented design changes across thousands of drawings. For example, the Employer’s engineers documented design revisions with handwritten notes, explicitly stating that modifications were required due to scope changes or previous errors, a critical piece of evidence, provided it is identified and properly analysed.
- Using geolocation and timestamps to link site photographs with specific project activities. As an example, AI analysed 25,000 site images, matched cable labelling to drawing revisions, and provided irrefutable proof that design modifications were necessary due to prior errors.
The human factor: AI as an assistant, not a replacement
A common concern: “Will AI replace experts?” The answer is a resounding no. While AI and data analytics significantly enhance dispute resolution, expert judgment remains critical.
AI tools provide efficiency, but human expertise ensures:
- Proper contextual interpretation of extracted data.
- Strategic questioning and legal positioning.
- Clear, evidence-based report writing tailored to dispute resolution forums.
AI is not a substitute for professional expertise, it is a tool that allows experts to focus on high-value analysis rather than sifting through massive data sets.
"Most people dream of retiring at 60. If you manually review every document, you'll be lucky to retire by 120... On Monday, you're optimistic. On Tuesday, you're overwhelmed. By Wednesday, you are questioning your career choices."
Lupa CTO, Djordje Nedeljkovic.
Conclusion: the future of construction dispute resolution
The industry is evolving. Traditional methods of dispute resolution relying on manual review, selective documentation, and subjective argumentation are no longer sustainable. The new way – leveraging automation, transparency, and AI-driven workflows coupled with construction specific advanced analytics can help to ensure disputes are resolved fairly, efficiently, and based on facts.
By integrating AI-driven analytics and data visualisation, the dispute resolution process can become:
- Faster, reducing document review from months to hours.
- More transparent, ensuring decisions are based on comprehensive, factual evidence.
- More reliable, eliminating subjective biases and strengthening claims.
As the construction industry evolves, the ability to harness data effectively will determine the success of claims and the resolution of disputes. The key question is no longer if AI will be used in dispute resolution but how it will be leveraged to ensure accuracy, efficiency, and fairness in resolving complex construction and engineering disputes.
This article was originally written for issue 28 of the Diales Digest. You can view the publication here: www.diales.com/diales-digest-issue-28